Things to think about:

  • What is hard about this?
    • Making software that is usable for this
  • How will we know whether CommCare is succeeding
    • Health improvements

Make list of hypothesis

usability

  • training
  • error rates

improvements

  • in follow-up
  • better coverage, houses visited

health outcomes

  • abc

Why is this better?

  • better tool
  • ppl feel better about stuff
  • novelty effect
  • Talk to social scientists -- maybe Batya

Paper, CommCare?, paper + some useless tech.

Below are a few ideas for research items we could study during the july-september 2009 pilot of CommCare

Improved care

The ultimate goal is to help the CHWs provide better care.

Measurable out comes

  • Number of houses visited
  • Number of houses using bed nets
  • Number of houses screened for TB
  • Improved health outcomes
    • Lower child mortality if CHWs do a better job of screening for childhood illness
    • Likewise, fewer deaths from TB if we're doing a better job screening for TB.

Time management

With the CommCare tool, CHWs would be able to better manage there time both at the day level and at the household visit level.

Measurable outcomes

  • Number of houses visited per day
  • Amount of data collected per house
  • Number of houses visited per village

Experiments to run

  • With and without a running clock during household visit
    • Could go either way
  • Color schemes and ordering for day planning
    • Feels like a strawman
  • Could do something comparing different scheduling devices
    • CC to normal calendar/schedule app on phone to paper

Related work space to explore

  • Perhaps expert systems
  • Old HCI work, human factors
  • Computers telling humans what to do and humans maybe/maybe not doing them
  • Check with Dan Weld and/or

Data quality

  • UI techniques for getting better quality data

Longitudinal data

The CommCare application will provide the CHW with better access to longitudinal data than current systems.

Measurables

  • How many follow ups were performed out of how many should have been performed

Experiments

  • Track current system to see how many follow ups that should have been done actually were done (and tell when they weren’t for ethical reasons)
  • Measure same

Challenges

  • No idea how many follow ups we are talking about

Training

Training the CHW versus not and how it effects their ability to use the software and success using it.

Related work

  • Immediate usability
    • Ben Shneiderman has done some work with immediate usability for public access systems. There is a paper that extends the four stages of a public access system: attraction, learning, engagement, and disengagement.
    • In the paper they present guidelines that are barely useful (at best) for what we're doing. However, could imagine writing a similar paper, adapting the guidelines fo r how to create immediately usable systems for developing regions.

Charts

Would come a bit later, but there may be some interesting issues with looking at the HCI issues around this population using visual representations of data.

Related work

  • Learning theory
  • When do people pick up abstractions?